Twitch Will Train Amazon's AI on Your Streams by Default—Here's How It Works and How to Opt Out

On Wednesday, August 12, 2026, Twitch announced that it will begin using streamers' channel content — enabled by default — to train Amazon's generative AI models, sparking immediate and widespread backlash from the community .

What Twitch Announced

Twitch updated its account settings to add a toggle labeled "Allow your channel content to train generative AI content models at Amazon" . The setting is enabled by default for all users, meaning streamers are automatically opted in unless they manually disable it . Twitch framed this as "adding controls for generative AI training," rather than as a new data collection initiative . The announcement confirmed rumors that had circulated for weeks about Amazon's plans to train generative AI on Twitch content .

What Content Is Being Collected?

The opt-out covers a broad range of channel content, including:

• Live streams (broadcasts)
• Video on Demand (VOD) recordings
• Clips
• Stream chat messages
• Other text and images on the channel

Importantly, turning off generative AI training does not opt users out of Twitch and Amazon using their content for other purposes described in Twitch's Terms of Service .

How the Opt-Out Setting Works — Including the Streamer-Overrides-Viewer Dynamic

The opt-out mechanism has several nuances that caught many users off guard:

• Streamers can disable the toggle in their account settings to exclude their own channel content (streams, VODs, clips, chat, and channel text/images) from future Amazon AI training .
• Critical override: If a streamer opts out, their preference governs the use of chat messages in their channel — meaning that even if an individual viewer has not opted out, their chat messages in that channel will not be used for training because the streamer's opt-out overrides the viewer's default .
• The opt-out is prospective only. Twitch disclosed that Amazon has already been using Twitch content for AI training for years, and the new setting does not retroactively remove previously collected data .

To find the setting, navigate to the Security and Privacy tab of your Twitch user settings and uncheck the "Training for Generative AI" toggle .

Scale of Community Backlash

The backlash was swift, intense, and concentrated. Major outlets described it as "swift and concentrated backlash" . Streamers and viewers criticized:

• The opt-out-by-default design
• The buried location of the setting ("hidden" in user settings)
• The fact that Amazon had already been training on their content without prior explicit consent
• The lack of compensation for creators whose content was being used

The anger was amplified by the perception that Twitch was "quietly" announcing a significant policy change via its support X account rather than through a more prominent communication .

Precedent for This Policy

Twitch's Terms of Service had long contained a broad license granting the platform extensive rights over user content, which several sources noted already covered this use case . Separately, a company executive confirmed more than two years ago that Amazon was using Twitch content for AI training . This announcement was therefore not a new practice but rather the introduction of a formal opt-out mechanism for an existing one.

In the broader tech landscape, several other platforms have shifted to an opt-out model for AI training (e.g., Google updated its terms to disclose public data use for AI models; Meta informed European users of its training practices) , but the combination of default opt-in, buried settings, and retroactive training drew particular ire in Twitch's case.